Yes, this is a good idea, although I'd like to note there is more nuance.

* `encoding` is what vim uses internally -- the way buffers and registers are stored in memory. Default value is latin1 as you mention, but it will also pick up on the $LANG environment variable. It's more effective in general to set that variable than directly tweak `encoding`. (Mine is en_US.UTF-8 due to my locale.)

* `fileencoding` is the encoding for the current buffer on disk, how it will be read and written. It is translated in memory to `encoding`. Thus you can manipulate files in encodings other than the one used in-memory. (Although if encoding was latin1 I'm not sure how well the round-trip would go.)

* `termencoding` is how the buffer will be displayed to the screen and how the user input will be read. Once again, it can differ from encoding, and will be translated.

* `fileencodings` is an ordered list of options to try when opening a file. Default value: "ucs-bom,utf-8,default,latin1" when encoding is a unicode value.

If you were editing Japanese text, for instance, choosing UTF-16 encoding would actually make vim use less memory than UTF-8, and would not affect the fileencoding used on disk either.